Hi Harshal, Embedded Pig in Python is supported from Oozie 3.1.3. Also, that seems a better option than creating dependencies between actions.
Thanks, Virag On 9/27/12 9:40 AM, "Alan Gates" <[email protected]> wrote: >I'll let others answer the Oozie questions as they'll be better >qualified. But in Pig 0.9 you can integrate the Pig with Python to do >the for loops. See http://pig.apache.org/docs/r0.9.2/cont.html > >Alan. > >On Sep 27, 2012, at 6:49 AM, Harshal Vora wrote: > >> Hi, >> >> We have a pig script where we want to parametrize 'group by' clause. >> We have been using Oozie for all our jobs. We hope that depending on >> some input parameter, the pig script is executed 'n' number of times, >> every time with a new value for the grouping clause. The input and >> output folders will also depend on the grouping clause (This is not an >> issue since we can parametrize input and output in the workflow if we >> have the value for the grouping clause). >> >> We want to use only ONE coordinator job for this. >> >> >> Q) Does the latest oozie support embedded pig scripts (pig 0.9 onwards)? >> >> >> Q) If not, is it possible to somehow create a for loop by putting some >> logic in the workflow actions? >> >> For ex. >> Action A (java action) --> Action B (pig action), and also output's some >> state. >> Action B --> Action C (java action) >> >> Action C --> depending on the value outputted by Action A decides if it >> should execute action A again or END. >> >> If Action A is executed again, it will use the previously outputted >> state to decide the next state. >> >> >> Q) Is there any other better way to do this? >> >> >> Regards, >> >
